Great Expectations

(Originally published January 2015)

On January 18, 2015, the San Antonio Express –News published an editorial regarding Texas’ child protection system which asserted the following:


        Like no other state agency, Child Protective Services must navigate

        an impossible path. On the one hand, the expectation for CPS is that

        it should never fail children in broken homes. This is the right expectation.

        One child’s death despite CPS involvement is too many. And yet the tragic

        reality is CPS has failed many kids, year after year.


       This pressure plays out every day. If workers get too aggressive pulling kids

       from troubled homes, CPS is often accused of breaking up families. But if workers     

       miss key signs that foreshadow tragedies, the agency becomes a symbol of 

       incompetence and disregard for vulnerable kids.


The editorial goes on to describe a series of stories in the Austin American-Statesman,

whose reporters reviewed 800 child death reports between 2010 and 2014, and comments that “Nearly half of the children who died had some type of prior contact with CPS. Many came from families that had more than three CPS investigations. If such history doesn’t trigger alarm bells with CPS investigators, we can’t help but wonder, what would? (author’s emphasis)


This editorial incorporates two assumptions that are common in media indictments of child protection programs:


  1. If more vulnerable children were placed out of the home, there would be fewer child maltreatment deaths.

  2. If children who died due to abuse and neglect (and their families) had prior contact with CPS, then CPS incompetence was likely a factor in the child’s death.


Further, the editorial asserts that the “right expectation” for CPS programs is that
“it should never fail children in broken homes,” an expectation that implies that CPS programs should be expected to prevent all child maltreatment deaths of children

whose families have ever been investigated or assessed by the state’s CPS program.


Of course, given information regarding large numbers of child deaths possibly related to child maltreatment, it’s usually not difficult to find examples of CPS incompetence; but this is far different than finding that lack of conscientious or skilled CPS performance was a key factor in a majority of maltreatment deaths in which children had CPS histories, or that incompetent performance in a specific case was reflective

of the state’s CPS practice as a whole. The assumption that most or all child maltreatment fatalities reflect CPS incompetence  also ignores the reality that CPS systems in the U.S. have inherent limitations based on both state law and limited resources:


  • CPS interventions follow reports of alleged abuse or neglect, and these interventions are often too little too late for many seriously endangered children.

  • State laws provide explicit criteria that must be met for child removal. A history of prior CPS reports on a family, even several reports, is not in and of itself grounds for child removal, any more than are “alarm bells” or the presence in families of common risk factors such as substance abuse or domestic violence. Child welfare agencies must be able to demonstrate risk of imminent harm, substantial endangerment, etc. to legally justify coercive out-of-home placements.

  • The risk assessment tools CPS programs have at their disposal do a good job of risk estimation for large groups of children and families, but do a poor job of predicting outcomes in specific families; the newer predictive risk models that use multiple administrative data bases are likely to have the same strengths and deficiencies, especially in regard to a low base rate phenomenon such as child fatalities. Concretely, this means that CPS caseworkers and supervisors are unable to predict which children will die in abuse/neglect related incidents.

  • Due to workload pressure and resource limitations, CPS programs often target resources at children and families in substantiated cases of

  • child maltreatment, or at high risk families, while foregoing early interventions for lower risk families that have a higher probability of achieving positive results.


Nevertheless, the common sense understanding that CPS failure to remove children from dangerous homes has been a key factor in some child maltreatment related deaths makes it difficult for concerned citizens, child advocates and policymakers to grasp the limitations of child placement as a strategy for reducing the number of child maltreatment deaths. A recent story in City Limits regarding New York City’s child protection system quotes the director of the city’s public child welfare agency, Gladys Carrion: “There is no correlation between the number of child fatalities and the number of removals. I don’t know why, I wish I knew.” For example, according to the reporter, Rachel Blustain, 13,345 children were placed out of the home in New York City in 1997, a year in which there were 43 child deaths of children “previously known to the system.”

In 2010, after several years of efforts to reduce New York City’s foster care population,

7108 children entered out-of-home care; in 2010, there were 39 deaths of children previously known to the agency. In 2005, 4813 children entered out of home care in New York City and 30 children previously known to the agency died from possible abuse or neglect.  A year later, after a highly publicized child maltreatment death, 6285 children were placed in foster care, and there were 45 child deaths possibly due to abuse or neglect of children who had prior contact with the child welfare system.


Children are placed in foster care for many different reasons and stay in care for widely varying periods of time. Children who enter foster care during their school age years, 6-17, die in abuse neglect related incidents at rates far lower than rates for younger children, especially babies and toddlers. However, the City Limits analysis of the relationship between entries into care and child maltreatment deaths and the child welfare director’s puzzlement points to a stark statistical reality: the number of children with prior CPS history in New York City is very large, several times larger than the approximately 75,000 children (50,000 investigations annually x 1.75 children per family minus the number of children named in a second or third report during the year) who live in families with screened in CPS cases during a single year. Even when New York City’s child welfare system increased entries-into- care by a few thousand children in a single year or two, the number of children who lived in families investigated or assessed by CPS during the past decade was reduced by only a small percentage. Many of these families are likely to have had the same risk factors, i.e., substance abuse, depression, DV, severe poverty, as families whose children were removed from the home. The pool of families with prior CPS history (over a decade or more) is very large, and the factors that increase the vulnerability of children in these families are common. Furthermore, the number of entries-into-care in a state or large city is likely to have little no influence on the vast majority of families whose child welfare cases may have been closed for months or years.


The same type of analysis can be applied to the children served by a child welfare system in a single year. Washington State’s Children’s  Administration (CA) opens cases on approximately 50,000 (unduplicated) children per year, and places about 5,000 of these children in out-of-home care due to child protection issues. Perhaps one-fifth of these children remain in care for only a few days or weeks leaving about 4000 children in foster care or with unlicensed relatives for significant periods of time. Even if CA doubled its placement rate, almost certainly a legal and practical impossibility, the number of children with open CPS cases at some point during the year living in birth families would be many times greater than the number of children placed in foster care for extended periods of time. Furthermore, many of these children’s parents and other caregivers are likely to have the same risk factors that characterize caregivers whose children have been placed in foster care. Absent dramatically improved risk assessment tools or predictive risk models, an increase in entry-into-care rates is unlikely to have much effect on numbers of child maltreatment deaths of children with open CPS cases in a single year, even if most children placed out of the home have been classified as high risk or been substantiated for child maltreatment. 


The limits of accountability


During 2012- 2013, Texas’ public child welfare agency opened approximately 165,000

Investigations per year, cases which may have involved possibly  250,000 (unduplicated) children. Over a period of 5 years (2010-14), Texas’ CPS system may have opened cases on several hundred thousand additional children assuming that children in families with multiple reports are counted only once. Is it reasonable to hold

Texas’ CPS system responsible for the maltreatment death of any child whose family had an open CPS case during these years, regardless of the time elapsed between case closure and the child’s death? Surely, the answer is “no”.  Nevertheless, this is the implicit standard frequently applied in media analyses of large samples of child deaths suspected to have been related to child maltreatment. These analyses may serve as useful reminders of the limitations of CPS programs in preventing severe injuries or child deaths, but not of CPS “incompetence”, absent detailed accounts of specific cases.


In my view, it makes sense for child welfare agencies to track suspected child maltreatment cases on open or recently open (in the past 12 months) CPS cases, in part to provide an incentive against premature case closure. It is also important that public child welfare agencies be held strictly accountable for honest, accurate and transparent classification and counts of suspected child maltreatment deaths. Currently, it’s common for child welfare agencies to use arbitrary indefensible criteria for classifying and counting child deaths, for example never classifying a child death as neglect related unless a coroner or medical examiner has made this determination ( a rare event in my experience) or unless a parent has been criminally charged ( equally rare in neglect related deaths). Child welfare agencies can delay classifying suspected cases for months or even years while fatality review processes occur. Agencies can change their criteria for counting in ways that reduce the number of child deaths reported in NCANDS and make it impossible for well-informed critics or advocates to track trends over a period of several years. These practices should be viewed as malfeasance in that as long as they are common within state systems, it will be impossible to know if child maltreatment deaths are increasing, declining or remaining stable in various states or at the national level. However, even when state child welfare systems are transparent regarding how they classify and count child maltreatment deaths, it is difficult to track trends in maltreatment deaths or compare states’ child death rates due to the varying definitions of “neglect related death” utilized by states and other rules/ policies applied in case counts.     


What should states and the federal government do to reduce child maltreatment deaths?


The importance of consistent, accurate, transparent measurements of child maltreatment deaths cannot be emphasized too strongly. However, at the practice level what is essential is earlier interventions with several groups of vulnerable children and families, both prior to CPS reports and following screened-in reports. Concretely, this means that both state legislatures and the federal government must rebuild public health agencies that were greatly weakened in many states by budget cuts during the recent recession, and that these public health agencies be staffed and funded to offer services to the following groups of families prior to and immediately following CPS reports:


  • Any pregnant woman or mother of an infant receiving publicly funded substance abuse treatment or mental health services should be offered a range of voluntary family support services.

  • Any low income single parent should be eligible for publicly funded high quality child care and respite care.

  • Any family with more than one police report of domestic violence should receive crisis intervention services.

  • Both state and federal governments should make a commitment to ending child homelessness with a range of housing services and other supportive services described in a recent Sounding Board.


In addition, public health agencies should be funded to implement Safe Sleep initiatives and initiatives developed to help parents with the inconsolable crying of infants. Further, both state legislatures and advocacy groups should act to stop the physical punishment of infants and toddlers, 0-2. Pediatricians and child development specialists need to speak up persistently against hitting babies for any reason.


What CPS should do to reduce child maltreatment deaths?


Currently, it is difficult to imagine a more maladroit set of guidelines for reducing life endangering injuries to young children than the guidelines for intervention that have grown out of the conceptually flawed Safe vs. Unsafe dichotomy. Specifically, waiting for babies and toddlers to be assessed as ‘in danger’ before intervening, or engaging in safety plans is a formula for failure. CPS agencies need to be intervening earlier, before these young children are in danger. CPS agencies should be developing safety plans with any family assessed as at risk for child maltreatment in which parents or other caregivers have substance abuse or chronic mental health issues. CPS agencies need to invest hugely increased resources in implementation of in- home safety plans for children, 0-5.


Further, using risk assessment tools to target high risk families is not a thoughtful strategy for making use of limited resources. In many instances, low and moderate risk families are much more likely to benefit from available services than high risk families.


Finally, insisting that public systems should continually shop for programs off of lists of programs labeled as “evidenced based” or “promising” by authoritative entities or local experts is a misguided policy. What is needed is for state systems to build an infrastructure for the ongoing testing of programs, and to engage in a vigorous never ending search for evidence of effectiveness of key programs, especially programs for abusive or neglectful substance abusing families and families with recurrent interpersonal violence.




Austin American- Statesman, Missed Signs, Fatal Consequences: How Texas missed deadly patterns and key pieces of information that could have helped protect vulnerable children.”  January 11-13, 2015.


Blustain, Rachel, City Limits, Limits of Protection: Can Mayor’s Push Reduce    Child Abuse Deaths?  December 16-18, 2014.


Child Maltreatment 2012 and 2013, Administration for Children and Families, U.S. Department of Health and Human Services.

San Antonio Express- News, CPS does little to earn public trust, January 18, 2015.                            

© 2020 by Dee Wilson Consulting. Proudly created with